Traffic Sign Detection Based on SSD Combined with Receptive Field Module and Path Aggregation Network

Area of Science:

• Computer Vision
• Artificial Intelligence
• Machine Learning

Background:

• Traditional traffic sign detection struggles with complex scenarios in intelligent transportation and advanced driver-assistance systems.
• Existing methods face limitations in achieving high detection accuracy and efficiency.

Purpose of the Study:

• To propose an improved traffic sign detection algorithm addressing limitations in accuracy and efficiency.
• To enhance the detection of small traffic signs in challenging environments.

Main Methods:

• Developed the SSD-RP algorithm, combining Single Shot Multibox Detector (SSD) with Receptive Field Module (RFM) and Path Aggregation Network (PAN).
• Integrated RFM to enhance feature map receptive fields and semantics, improving small sign detection.
• Utilized PAN for multiscale feature integration, enhancing feature discrimination and location/classification accuracy.
• Incorporated spatial pyramid pooling to supplement fine-grained features.

Main Results:

• SSD-RP demonstrated higher mean average precision (mAP) on GTSDB and CCTSDB datasets compared to the traditional SSD algorithm.
• The proposed algorithm showed superior performance in detecting small traffic signs.
• SSD-RP achieved a better balance between detection time and precision than Faster R-CNN, RetinaNet, and YOLOv3.

Conclusions:

• The SSD-RP algorithm offers significant improvements in traffic sign detection accuracy and efficiency.
• It is well-suited for intelligent transportation systems and advanced driver-assistance systems.
• SSD-RP provides a competitive solution for real-time, high-precision object detection tasks.